1. Arnold School of Public Health Health Promotion, Education, and Behavior SOCIAL MEDIA MESSAGE TYPE, ENGAGEMENT, AND WEIGHT CHANGE IN A 6-MONTH BEHAVIORAL WEIGHT LOSS INTERVENTION Sarah Hales, LMSW Charis Davidson, MPH Brie Turner-McGrievy, PhD

2.  To examine  Whether different types of social network messages differentially affect participant engagement  If engagement with social media enhances weight loss as part of a 6-month behavioral weight loss intervention OBJECTIVE

3. THE NEW DIETS TRIAL The New Dietary Interventions to Enhance the Treatments for weight loss (New DIETs) study: 2-month weight loss intervention, with An optional 4-month follow-up period.

4. INTERVENTION: MONTHS 0-2  Vegan, vegetarian, pesco-veg, and semi- veg: Weekly group meetings (8 total)  Omnivorous group: Monthly meetings (3 total) Weekly newsletters Weekly e-mails All groups: Orientation and overview of diets in group meeting Diet information handouts Recipe books

5. INTERVENTION: MONTHS 3-6 All groups, including the omnivorous group, met monthly face-to-face. All groups were provided with a private Facebook group for social support in between meetings.

6.  Data are a sub-analysis from a 2-month randomized weight loss study with 4-month follow-up support provided via private Facebook groups and monthly meetings.  Counselors posted 5 different message types/week based on Social Cognitive Theory (similar to format of group classes).  Order of messages each week was random except for weight messages (posted each Monday). METHODS

7. Message TypesTargeted Social Cognitive Theory Construct Weight LossSelf-control RecipeBehavioral capability Observational Learning (when links to videos of cooking demos were provided) Nutrition Study, Science, or News Situation PollTargeted various constructs but was included to mirror the questions counselors would ask of participants during face-to-face meetings Suggestion/RequestExpectations Emotional coping response FACEBOOK MESSAGE TYPES TARGETING SCT CONSTRUCTS

8. Message TypesMessage Examples Weight Loss“How has your weight loss been since last week? Up, down, or about the same?” Recipe“This makes for a great snack, sandwich spread, or pasta topping: (Link to recipe) Nutrition Study, Science, or News “Here is a great list of plant-based sources of calcium. Ever added kale to a smoothie? (link to website)” Poll“What’s the most challenging meal for you to prepare each day? Breakfast, Lunch, Dinner, or Snacks.” Suggestion/ Request “Share with the group and help others. How have you dealt with a weight plateau (or gain) recently?” FACEBOOK MESSAGE EXAMPLES

9.  Responses to counselor-posted messages:  Views  Likes  Comments and poll votes  Engagement was dichotomized using a median split (of total likes, comments, and poll votes over course of the study) MEASURING ENGAGEMENT

10.  Baseline, 2 months, 6 months:  Height (SECA 213, using a calibrated stadiometer)  Weight (SECA 869, Hamburg, Germany, calibrated digital scale accurate to 0.01 kg)  Facebook message preference  Assessed in the survey administered at 6 months OTHER MEASURES

11. Joined, not engaged Joined, engaged Did not joinTotal P-value for difference between groups n15 3363 Meetings attended in 4-month follow-up 1.7±1.43.7±0.8 b 1.8±1.42.3±1.5P<0.001  Participants were mostly highly educated (94% college or higher), white (79%), females (73%), with a mean age of 48.5 ± 8.3 years. DEMOGRAPHICS

12. WHICH MESSAGE TYPE PROMPTED THE MOST USER ENGAGEMENT?

13. MEAN NUMBER OF COMMENTS AND POLL VOTES PER POST IN RESPONSE TO DIFFERENT TYPES OF COUNSELOR FACEBOOK MESSAGES Sig greater than nutr news & recipes, P<0.05 F=60.3, P<0.001

14. MEAN NUMBER OF LIKES TO DIFFERENT TYPES OF COUNSELOR FACEBOOK MESSAGES Sig greater than weight & recipe, P<0.05 F=20.6, P<0.001

15. EXPOSURE TO DIFFERENT TYPES OF COUNSELOR FACEBOOK MESSAGES P=0.67

16.  Exposure was similar for all message types.  Posts asking participants to vote in a poll or request suggestions for the group are the most engaging as measured by both comments/poll votes and likes. RESULTS OF MESSAGE TYPE AND ENGAGEMENT

17. WHICH MESSAGES DID PARTICIPANTS REPORT LIKING THE MOST AT THE END OF THE STUDY? No one reported liking the weight posts.

18.  Facebook engagement was significantly associated with weight loss during the 4-month maintenance period (B= -0.09, P=0.04) such that for every 10 posts, comments, poll votes, or likes to Facebook, participants lost a mean of 0.43 kg. WAS ENGAGEMENT WITH FACEBOOK ASSOCIATED WITH WEIGHT LOSS?

19. FACEBOOK ENGAGEMENT AND WEIGHT LOSS  Facebook engagement was significantly associated with weight loss during the 4-month maintenance period (B= -0.09, P=0.04) such that for every 10 posts, comments, poll votes, or likes to Facebook, participants lost a mean of 0.43 kg. -0.4kg 10 posts!

20. WEIGHT LOSS BY 3 ENGAGEMENT LEVELS AT 2 MONTHS P=0.90 Adjusted for meeting attendance Pounds lost

21. WEIGHT LOSS BY 3 ENGAGEMENT LEVELS AT 6 MONTHS Sig diff from Didn’t join and Joined, engaged P<0.01 Pounds lost

22.  Messages soliciting feedback, such as polling features, prompt the most engagement when counselors post messages in a social network during a weight loss trial.  Engagement with social media was associated with greater weight loss during the maintenance period.  The findings provide evidence for ways to provide social support during weight loss interventions using remote methodology. CONCLUSIONS